Use cases and roles

Root Cause Assist is designed to evaluate and summarize insights from your existing reporting dataset, specifically the intersection of metrics, Text Analytics, sentiment, and segmentations across your organizational hierarchy. Align your use cases with these core elements so that Root Cause Assist generates actionable, high impact insights for your organization.

Prioritize use cases and roles where analyzing trends across aggregated datasets is a core need as Root Cause Assist is designed to highlight top and bottom trending data elements (e.g. metrics, topics, sentiment, segments, unit groups).

Ensure that each role has access to a representative dataset, typically 100 records or more for the time period, so that Root Cause Assist has a sufficient amount of data points to flag potential anomalies. If there is an insufficient dataset, or if the above core elements are not present in the dataset, you may receive an error message.

Text Analytics configuration

Text Analytics is a mandatory parameter for the Root Cause Assist feature to function as unstructured text data including survey feedback, transcripts, and social reviews contextualize the underlying quantitative drivers (e.g. metric, score change, impact score, net sentiment) evaluated by Root Cause Assist. In order to generate actionable insights, only use Root Cause Assist on records where Text Analytics processing has been completed.

Ensure that your Text Analytics program provides comprehensive coverage for all key channels, signals and use cases. We strongly recommend that you leverage level 2 or lower in your Text Analytics hierarchy in order to provide more granular insights related to unstructured text data.

Segmentation

Root Cause Assist is designed to pinpoint specific populations of interest in its summary, meaning that effective segmentation across your dataset is essential to uncovering more granular insights. For each use case and channel, confirm that your reporting configuration captures meaningful customer segmentations that can be used for insights discovery. Some examples may include interaction channel, customer loyalty tier, contact reason, or product type.

Be sure to select all relevant segments for each role that you would like Root Cause Assist to evaluate when setting up shared configurations in Admin Suite. Typically each role should have 3-5 relevant segments of interest in order for Root Cause Assist to generate contextual, multi-dimensional analyses.

Organizational hierarchy and reporting settings

Aligning your Root Cause Assist deployment with the appropriate level of the organizational hierarchy is a critical component of success as it ensures the analysis spans the proper breadth and depth for the needs of each role. For example, a company executive or insights role would best be served by analyzing performance across multiple regions and/or business units while frontline roles may want a lower level view of regions, locations, and teams.

When selecting unit groups in shared configuration for use by Root Cause Assist, consider not only the appropriate level of the organizational hierarchy that aligns with the role(s) but also the number of units in the unit group. Root Cause Assist is designed to evaluate the selected organizational hierarchy unit groups and highlight interesting data points and anomalies, meaning the pool of eligible units should be right sized—not too large or too small. Typically unit group evaluation should be aligned with anywhere from 10-100 individual values for the best results. If the number of units is too large, consider analysis with a higher level in the organizational hierarchy (e.g. regional view vs. store level).

Setting an appropriate minimum sample size in addition to proper unit group alignment is another means of tailoring the Root Cause Assist configuration to maximize the value and relevance of summarized insights. As a general rule of thumb, the higher the alignment with organizational hierarchy (e.g. executive, insights) the higher a minimum sample size threshold should be set, and vice versa. The best practice recommendation for minimum sample size is anywhere from 100-1000 for company level roles and 1-100 for frontline roles, but this is highly dependent on the size of your organization and volume of incoming signals being analyzed.

Recommended configuration by scope of analysis

Use case examples

The following use cases provide a blueprint for designing a successful Root Cause Assist deployment. These examples illustrate how to align program design with the unique needs of various roles within an organization.

Use case 1: Retail - regional manager

As a regional manager for a large electronics retail chain with 250 nationwide locations, I oversee the Northeast region. My key responsibilities include the performance of my stores, measured by metrics such as NPS and CSAT.

To improve both regional and store performance, I rely on Root Cause Assist to identify emerging trends that negatively impact customer satisfaction. These trends could include issues like return policies, staff helpfulness, item availability, pricing, or promotional offers.

It's crucial for me to have granular insights, understanding which specific stores and customer segments are most affected. This level of detail, including the ability to drill down into individual customer comments, allows me to develop targeted, actionable next steps and truly empathize with our customers.

To ensure Root Cause Assist provides the most relevant and valuable insights for my role, which requires a medium to lower level of granularity, I need to collaborate with my Medallia administrator. This ensures the proper configuration of Text Analytics topics, themes, customer segments, and stores for evaluation. Additionally, I will set a minimum sample size of 100 records to filter out low-volume feedback trends that are not actionable or won't significantly impact my regional KPIs.

Use case 2: Retail - CX executive

As a CX Executive, I am accountable for the company's overall performance in customer satisfaction, revenue, and brand loyalty. When using Root Cause Assist, I need visibility into regional and company-level NPS and CSAT trends, focusing on high-level themes that directly influence our business KPIs. Understanding drivers and shifts in customer sentiment is also critical for my evaluation.

My Medallia administrator will configure Root Cause Assist to provide a top-down, company-level view, broken down by region. Store-level information is too granular for my role, as I delegate those insights to Regional Managers. My focus is on macro trends, meaning high-volume occurrences are most important for my evaluation and a typical minimum sample size of 1000 records is the best practice. I also need to understand score impacts by customer loyalty tier to drive strategic improvement programs that maximize our business ROI.

Use case 3: Financial services - insights

As an insights manager for a financial services institution, I am responsible for identifying key friction points in the customer journey and delivering recommendations to various stakeholders (e.g. executives, regional managers) to ensure we maintain high customer satisfaction and seamless experiences across all customer touchpoints.

Root Cause Assist enables me to understand trending issues that affect customer satisfaction and increase operational costs. I may find that long wait times at branches, limited in-branch service offerings, or mobile app crash issues are not only driving up contact center volumes, but also eroding NPS and brand loyalty, and leading to customer churn. It may also highlight a specific branch region or contact center manager who is under-performing and could use active coaching.

In order to get the most out of Root Cause Assist, I have configured an omnichannel view of Text Analytics topics and themes across branch locations, digital channels, and contact center interactions with level 2 granularity to obtain specific, actionable insights.

In addition, my selection of segments spans customer demographics, product types, and loyalty tiers, along with a regional-level unit group view, provides the right level of granularity to identify high impact issues affecting various cohorts across the business. To filter out the noise and focus on the highest ROI opportunities, I set my minimum sample size to 250 records to ensure I deliver pertinent high value insights for my organization.

Use case 4: Hospitality - property manager

As a property manager for a global hospitality brand, I am motivated and incentivized to deliver top-notch experiences for our guests. These experiences are best indicated by high NPS, LTR, and CSAT scores, as well as driver questions such as cleanliness, amenities, and professionalism. Root Cause Assist enables me to get a quick pulse check as to my property’s performance and, if our key scores or drivers are trending in a negative direction, what is causing issues for our guests and how to address them.

In order to be successful with Root Cause Assist, I asked my Medallia administrator to include all relevant information Text Analytics topics and themes that would be relevant to the on-property guest experience as well as digital tools such as booking experience and the mobile app. It’s also important for me to understand how we perform across different customer segments such as loyalty tier, purpose of visit, first-time vs. repeat visitor, and guest demographics. When it comes to feedback, every soundbite matters; therefore, the minimum sample size for Root Cause Assist is set to 1, as no issue is ever too big or too small to drive great experiences at my property.

Conceptual diagram for Root Cause Assist configuration

Root Cause Assist leverages the concept of shared configuration to enable tailored role-level configuration of data entities, segmentations, and settings which are used to produce Root Cause Assist generative AI summaries. The preceding use cases indicate common use cases and best practices for setting up shared configuration, which are essential for success when using Root Cause Assist. The following diagram depicts how these concepts are leveraged in the system workflow.